Multi-domain Fake News Detection for History News Environment Perception

As social networks continue to grow, fake news in different areas on social media is spreading widely. Numerous parts of the actual world, such as politics and daily living, are threatened by fake news. Most traditional fake news detection focuses on a single domain, so identifying multi-domain fake news is receiving increasing attention. To improve the detection of multi-domain fake news, a novel multi-domain fake news detection model is proposed. We created the History News Environment Perception framework by analyzing the relationship between domain news and its news environment sample (HNEP). The approach is that for each post, we build the history and mainstream news environment from all previous posts. We design a perception module for mainstream and history news environments to improve multi-domain fake news prediction by fusing perceptual representations via domain gates. The experimental results demonstrate that HNEP can enhance the performance of multi-domain fake news detection.